The effect of voluntary disclosure on stock liquidity: New evidence from index funds

Highlights

• Since voluntary disclosure can affect information asymmetry among traders, theoretical research has shown that voluntary disclosure can affect stock liquidity.

• This study empirically tests whether voluntary disclosure affects stock liquidity.

• When a firm joins the S&P 500 index, voluntary disclosure increases with the level of ownership assumed by index funds.

• This increase in disclosure is associated with increased stock liquidity, which implies that voluntary disclosure increases stock liquidity.

Abstract

This study tests whether voluntary disclosure affects stock liquidity. I argue that index funds fit the profile of nonstrategic traders who, according to theory, are unambiguously more likely than managers and strategic investors to prefer high stock liquidity and thus high disclosure. This suggests that I can use index funds' disclosure preferences to construct an empirical model of voluntary disclosure that abstracts away from managers' strategic disclosure motives. Accordingly, I use an index-fund setting to construct a recursive structural equation model of voluntary disclosure, index fund ownership, and liquidity. I find that when a firm joins the S&P 500 index, voluntary disclosure increases with the level of ownership assumed by index funds, and this increase in disclosure is associated with increased stock liquidity. These results imply that voluntary disclosure increases stock liquidity.

Introduction

Stock liquidity is central to the efficient functioning of trade in the financial markets and is determined in part by information asymmetry among traders. Since voluntary disclosure can affect information asymmetry among traders, theoretical research has shown that voluntary disclosure can affect stock liquidity (e.g., Bushman and Indjejikian, 1995). However, empirical tests of this mechanism must recognize that a manager's provision of disclosure is a product of many forces, all of which must be identified.¹ It has proved difficult to construct an empirical model of disclosure that incorporates all these forces (e.g., Joos, 2000, Leuz and Verrecchia, 2000). Index funds, by contrast, trade primarily for nonstrategic reasons, and theory suggests that such nonstrategic traders unambiguously prefer higher disclosure relative to privately informed managers and strategic investors. This implies that I can use index funds' disclosure preferences to construct an empirical model of voluntary disclosure that abstracts away from managers' strategic disclosure motives. Accordingly, I use an index-fund setting to construct a recursive model of voluntary disclosure, index fund ownership, and stock liquidity, and demonstrate that voluntary disclosure increases stock liquidity.

The effect of index fund ownership on liquidity can arise from two mechanisms: (1) index funds' unambiguous preference for additional disclosure, and (2) index fund ownership itself (e.g., increased nonstrategic trading related to portfolio rebalancing). Separating these two mechanisms is crucial for understanding the determinants of liquidity, but it cannot be done with simple OLS or regression discontinuity (RD) techniques. For example, even if index funds elicit more disclosure from managers, instrumenting disclosure with index fund ownership does not solve the identification problem: Any effect on liquidity related to common variation in disclosure and index fund ownership could be due to either increased disclosure or index fund ownership itself (the same is true for RD designs). An appropriate empirical solution for separating these effects is to construct structural equations for disclosure and index fund ownership, and then to use recursive estimation to obtain each mechanism's liquidity effect (Greene, 2002, p. 397).²

Constructing the recursive model requires that index fund ownership be properly identified. To achieve this goal, I use an index-fund-inclusion setting. An important feature of this setting is that changes in the index's constituent firms are made by the index's governing body (such as S&P), and are therefore exogenous to index funds themselves. When a firm is added, index funds must take an ownership position in that firm, and this exogenous ownership position confers control rights. When index funds acquire control rights through ownership, any link between an increase in voluntary disclosure due to this control and stock liquidity can be construed as causal. By contrast, it is harder to make such claims for strategic investors (including managers) without explicitly modeling their endogenous preferences for disclosure, ownership, and liquidity. The same goes for atomistic or retail traders, who can trade at will according to their disclosure preferences and are too small to influence managers' disclosure choices.

I use inclusion in the S&P 500 index as the key economic phenomenon to test whether disclosure affects stock liquidity. At the time of a firm's index inclusion, the ownership level assumed by index funds is determined by their assets under management (AUM). Since index fund AUM varies across time and index inclusions occur throughout time, the index-fund setting generates substantial variation in index fund ownership. This setting has other benefits as well. Since firm managers can neither influence index fund AUM nor control the timing of index inclusions, the index fund ownership stake is exogenous to the firm (see Section 3.1). From the index fund perspective, prior research suggests that variation in index fund ownership is not due to strategic stock selection (see Section 2). In addition, S&P makes public their index-inclusion criteria, which I use to match each inclusion firm to a control firm (using propensity scores). Exploiting each inclusion firm's behavior pre and post inclusion relative to the matched control firm results in a difference-in-differences (D-in-D) regression setting that eliminates any time-varying effects common to both firms and differences away any firm-fixed effects. Importantly, the D-in-D design also controls for any “index-inclusion” effect common to all inclusion firms (for disclosure and liquidity).

I use a sample of 368 firms added to the S&P 500 index over the period of 1996–2010 (and 368 control firms). In my sample, 78% of index inclusions occur due to a merger or an acquisition involving an existing index firm, which frees up a spot in the index. Upon index inclusion, the ensuing index fund ownership level in my sample firms is economically meaningful, ranging from 3.9% to 10.2% of a firm's common shares outstanding.

In the first analysis, I regress several disclosure proxies on the change in index fund ownership resulting from index inclusion. I find that for a one standard deviation increase in index fund ownership, management guidance disclosures increase by approximately 19.35%, 8-K filings increase by 18.99%, supplementary financial statement filings increase by 9.49%, and press releases increase by 7.51% (relative to control firms). I also find that less entrenched managers provide more disclosure for the same amount of increase in index fund shareholdings. These results suggest that index funds are more effective at eliciting disclosure when they have a larger stake in a firm. Inclusion firms also experience an improvement in their information environment, as measured by a reduction in analyst forecast errors.³

In the second analysis, I demonstrate that stock liquidity, as proxied for by bid-ask spreads and the Amihud (2002) illiquidity measure, increases for inclusion firms relative to control firms in a D-in-D manner when the inclusion firm gets a larger ownership stake from index funds. However, recall that this increase in liquidity may be unrelated to the improved disclosure environment—it could occur simply because of index fund ownership by itself. I use a recursive structural equation model to separate the disclosure effect and the index fund ownership effect on liquidity, and find that the disclosure effect is significant (see Section 4.5). This result implies that disclosure increases liquidity.

This study makes several contributions to the literature. First, other studies relating disclosure to stock liquidity acknowledge identification challenges in modeling managers' strategic disclosure choices. For example, Leuz and Verrecchia (2000) rely on the voluntary adoption of IFRS by German firms to represent a commitment to increased disclosure and find that voluntary adopters experience greater liquidity. Shroff et al. (2013) find that after the passage of the 2005 Securities Offering Reform, firms increased disclosure before seasoned equity offerings, which led to higher liquidity for these firms. Balakrishnan et al. (2014) find that increases in disclosure associated with reductions to analyst coverage lead to increased liquidity, but they do not explicitly model the increase in disclosure. My study corroborates the prediction that disclosure increases liquidity, yet it abstracts away from any particular model of managers' strategic disclosure motives, which considerably simplifies the analysis. I also highlight an important mechanism that gives rise to disclosure.

Second, through the use of index funds, I bypass an endogeneity concern that is specific to institutional holdings. Prior studies such as Lang and Lundholm (1996) and Bushee and Noe (2000) show that disclosure is associated with institutional investor shareholdings. These studies use these associations to argue that institutional investors are attracted to firms with certain disclosure practices, and that managers adopt disclosure practices to attract such investors. However, these studies acknowledge that institutional investors may strategically invest and divest based on their unobservable investing strategy, which may ultimately affect their demand for disclosure. Index funds, by contrast, have no such leeway in choosing their portfolio firms. Instead, the funds appear to influence disclosure through ownership, an important consideration for future work on institutions and disclosure.

Third, my study differs in two important ways from that of Boone and White (2015), who argue (along lines similar to mine) that index funds prefer high disclosure because it decreases trading costs. Boone and White (2015, Tables 3 and 5) show that firms moving from the Russell 1000 to the Russell 2000 index exhibit increased disclosure, increased liquidity, and higher institutional ownership. They acknowledge that the liquidity effect could occur “via direct and indirect channels,” with the former related to index fund ownership and the latter related to disclosure. But, unlike this study, theirs does not explicitly test for the two mechanisms and thus cannot address the economic question of whether disclosure and index fund ownership separately affect liquidity. Another difference is that Boone and White (2015) measure institutional ownership at the “quasi-indexer” level. Bushee and Noe (2000) find that this classification comprises not just index funds but also non-index funds that have strategic disclosure preferences. By measuring index fund ownership directly from regulatory filings, I eliminate the possibility that institutions' strategic disclosure preferences drive my results.⁴

Lastly, my focus on index funds complements the work of Brav et al. (2008), who show that other large owners, such as hedge funds, exercise their ownership power by intervening in managers' business decisions. Such intervention, which requires the investor to acquire new management skills, is not feasible on a large scale for index funds, which aim to hold a large basket of stocks in many industries at low cost. Nonetheless, as this study shows, index funds do exploit their ownership power for other purposes.

In Section 2, I motivate the hypotheses. 3 Data, 4 Empirical results describe the data and empirical results, and Section 5 concludes.